boosting原理

weight D(1)

training

Training set(1) with weight D(1)

update weight

weight D(T)

Weak learning classifier(T)

Weak learning classifier(2)

weight D(2)

Weak learning classifier(1)

update weight

Strong learningclassifier